Header

UZH-Logo

Maintenance Infos

Z-Tests in multinomial probit models under simulated maximum likelihood estimation: some small sample properties


Ziegler, Andreas (2010). Z-Tests in multinomial probit models under simulated maximum likelihood estimation: some small sample properties. Jahrbücher für Nationalökonomie und Statistik:630-652.

Abstract

This paper analyzes small sample properties of several versions of z-tests in multinomial probit
models under simulated maximum likelihood estimation. OurMonte Carlo experiments show
that z-tests on utility function coefficients provide more robust results than z-tests on variance
covariance parameters. As expected, both the number of observations and the number of random
draws in the incorporatedGeweke-Hajivassiliou-Keane (GHK) simulator have on average
a positive impact on the conformities between the shares of type I errors and the nominal significance
levels. Furthermore, an increase of the number of observations leads to an expected
decrease of the shares of type II errors, whereas the number of random draws in the GHK
simulator surprisingly has no significant effect in this respect. One main result of our study
is that the use of the robust version of the simulated z-test statistics is not systematically
more favorable than the use of other versions. However, the application of the z-test statistics
that exclusively include the Hessian matrix of the simulated loglikelihood function to estimate
the information matrix often leads to substantial computational problems.

Abstract

This paper analyzes small sample properties of several versions of z-tests in multinomial probit
models under simulated maximum likelihood estimation. OurMonte Carlo experiments show
that z-tests on utility function coefficients provide more robust results than z-tests on variance
covariance parameters. As expected, both the number of observations and the number of random
draws in the incorporatedGeweke-Hajivassiliou-Keane (GHK) simulator have on average
a positive impact on the conformities between the shares of type I errors and the nominal significance
levels. Furthermore, an increase of the number of observations leads to an expected
decrease of the shares of type II errors, whereas the number of random draws in the GHK
simulator surprisingly has no significant effect in this respect. One main result of our study
is that the use of the robust version of the simulated z-test statistics is not systematically
more favorable than the use of other versions. However, the application of the z-test statistics
that exclusively include the Hessian matrix of the simulated loglikelihood function to estimate
the information matrix often leads to substantial computational problems.

Statistics

Downloads

140 downloads since deposited on 04 Mar 2011
23 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Journal Article, refereed, original work
Communities & Collections:03 Faculty of Economics > Center for Corporate Responsibility and Sustainability
Dewey Decimal Classification:330 Economics
Date:2010
Deposited On:04 Mar 2011 09:55
Last Modified:07 Apr 2017 14:24
Publisher:Lucius & Lucius
ISSN:0021-4027
Official URL:http://www.wiso-net.de/webcgi?START=A60&DOKV_DB=ZECO&DOKV_NO=JFNSB2CF2EF544EF37E5E738940C6EAD4213&DOKV_HS=0&PP=1

Download

Download PDF  'Z-Tests in multinomial probit models under simulated maximum likelihood estimation: some small sample properties'.
Preview
Content: Accepted Version
Filetype: PDF
Size: 361kB
Download PDF  'Z-Tests in multinomial probit models under simulated maximum likelihood estimation: some small sample properties'.
Preview
Content: Published Version
Filetype: PDF
Size: 1MB